pandas/pyarrow conflict on colab

[i-pj]

Current imports conflict with the colab's built in packages.

Error log:

ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
google-colab 1.0.0 requires pandas==2.2.2, but you have pandas 2.3.3 which is incompatible.
dask-cudf-cu12 25.6.0 requires pandas<2.2.4dev0,>=2.0, but you have pandas 2.3.3 which is incompatible.
cudf-cu12 25.6.0 requires pandas<2.2.4dev0,>=2.0, but you have pandas 2.3.3 which is incompatible.
cudf-cu12 25.6.0 requires pyarrow<20.0.0a0,>=14.0.0; platform_machine == "x86_64", but you have pyarrow 21.0.0 which is incompatible.
pylibcudf-cu12 25.6.0 requires pyarrow<20.0.0a0,>=14.0.0; platform_machine == "x86_64", but you have pyarrow 21.0.0 which is incompatible.

[Copilot]

Pull Request Overview

This PR fixes dependency conflicts in Google Colab by pinning pandas and pyarrow to compatible versions and reorganizing the pip install commands. The changes address version incompatibilities between Numerai dependencies and Colab's built-in packages.

• Pin pandas to 2.2.2 and pyarrow to 19.0.0 to match Colab's compatibility requirements
• Split the installation into separate commands for better dependency management
• Remove the -q (quiet) and --upgrade flags from install commands

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}